The role of feedback contingency in perceptual category learning.

Abstract

Feedback is highly contingent on behavior if it eventually becomes easy to predict, and weakly contingent on behavior if it remains difficult or impossible to predict even after learning is complete. Many studies have demonstrated that humans and nonhuman animals are highly sensitive to feedback contingency, but no known studies have examined how feedback contingency affects category learning, and current theories assign little or no importance to this variable. Two experiments examined the effects of contingency degradation on rule-based and information-integration category learning. In rule-based tasks, optimal accuracy is possible with a simple explicit rule, whereas optimal accuracy in information-integration tasks requires integrating information from two or more incommensurable perceptual dimensions. In both experiments, participants each learned rule-based or information-integration categories under either high or low levels of feedback contingency. The exact same stimuli were used in all four conditions and optimal accuracy was identical in every condition. Learning was good in both high-contingency conditions, but most participants showed little or no evidence of learning in either low-contingency condition. Possible causes of these effects are discussed, as well as their theoretical implications.

Due Process in Dual Process: Model‐Recovery Simulations of Decision‐Bound Strategy Analysis in Category Learning

“…In the context of our suggestion to use maximum likelihood, there is a certain irony in that fact that the most recent COVIS papers tend to use the Bayesian Information Criterion (e.g., Ashby & Vucovich, ; Spiering & Ashby, ), while earlier papers tended to use the Akaike Information Criterion (Ashby et al., ; Ell et al., ; Maddox, Bohil, & Ing, ; Maddox & Ing, ). This is because both information criteria attempt to correct for model complexity using the number of parameters and the BIC penalizes high‐parameter models more heavily than the AIC (Myung & Pitt, ).…”

“…However, despite all these critiques, proponents of COVIS still publish studies looking to find a dissociation between learning of rule‐based and information‐integration category structures (e.g., Ashby & Vucovich, ; Smith et al., , ) and strongly recommend these category structures as appropriate for studying category learning (Ashby & Valentin, in press). It is these most recent (and future) studies that the work presented here most targets.…”

“…In addition to the choice of model‐selection statistic (see above), a number of other things differ between papers. For example, Ashby and Vucovich () analyzed only the last 100 training trials of their experiment, while Spiering and Ashby () analyzed each of four 150‐trial blocks across their experiment. Casale et al.…”

On Fodor's First Law of the Nonexistence of Cognitive Science

“…Within my field, the psychology of concepts, this became apparent in the many studies of the classic debate between prototype and exemplar models, when it was eventually realized that group results did not necessarily represent the individuals making up the group. When models were fit to individual subjects, it was often found that some people were fit by prototype models, some by exemplar models, and some by no model (e.g., Malt, ; Smith, Murray, & Minda, ; see also a different comparison by Ashby & Vucovich, ). Variation is found in other conceptual behaviors.…”

Increased cognitive load enables unlearning in procedural category learning.

“…Contrast this with veridical feedback, in which negative feedback is typically accompanied by low response confidence. We recently showed that category learning is exquisitely sensitive to this feedback contingency (Ashby & Vucovich, in press). (2) The TANs must close the gate when RF is detected.…”